On Fri, 26 Sep 2014, Felipe Balbi wrote: > Hi Alan, > > it seems like newer versions of USB20CV will fail us on the MSC tests > unless we support READ_CAPACITY(16). How is that possible? The USB Mass-Storage Class spec doesn't require the device to support any particular set of SCSI commands. READ CAPACITY(16) isn't even part of the SCSI-2 spec, which is what g-mass-storage supports. > There is one detail though, looking at the sniffer, CBW comes as: > > 55 53 42 43 43 59 45 4C 00 00 00 00 00 00 0A 2F 00 00 > 00 00 00 00 00 80 00 00 00 00 00 00 00 > > so SCSI cmd opcode is set to 0x2f. Looking at scsi.h that's defined to > VERIFY: > > 89 #define WRITE_VERIFY 0x2e > 90 #define VERIFY 0x2f > 91 #define SEARCH_HIGH 0x30 Yes, that's right. g-mass-storage does implement VERIFY. > I suppose this would be a Mass Storage quirk ? Any idea how to implement > READ CAPACITY 16 ? (yeah, haven't read SCSI spec yet, doing it now). According to scsi.h, READ CAPACITY(16) is service action 0x10 of SERVICE ACTION IN, which is 0x9e. The draft SBC-3 standard from t10.org agrees. I don't see how anyone could get 0x2f out of that. > Any hints will be appreciated. We could implement READ CAPACITY(16) if necessary. But since we don't implement READ(16) or WRITE(16), it seems rather pointless. Are you certain you are interpreting the USBCV output correctly? Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
